Real estate construction monitoring method and device of TimesNet model and medium

Abstract

The embodiment of the invention discloses a real estate construction monitoring method and device based on a TimesNet model and a medium, and relates to the technical field of model prediction.The method comprises the steps that multi-dimensional construction time sequence data of a real estate construction project is obtained, feature extraction is conducted on the multi-dimensional construction time sequence data through a graph attention network, and a real estate construction project is obtained; multi-dimensional fusion construction features corresponding to the multi-dimensional construction data are determined; through a pre-trained TimesNet model, according to the multi-dimensional fusion construction features corresponding to the multi-dimensional construction data, determining multi-dimensional prediction construction data in a corresponding prediction time length, the multi-dimensional prediction construction data including prediction construction resource data and prediction construction progress data, the prediction construction prediction resource data comprises prediction material resource data and prediction human resource data; and carrying out risk monitoring according to the multi-dimensional prediction construction data, determining construction risk information of the real estate construction project, and carrying out risk management and control on the real estate construction project through the construction risk information.

Description

Technical Field

[0001] The present specification relates to the field of model prediction technology, and in particular to a TimesNet model real estate construction monitoring method, equipment and medium. Background Art

[0002] In the field of real estate construction project management, as the scale and complexity of projects continue to expand, comprehensive, real-time and accurate monitoring of construction data has become particularly important. Traditional construction monitoring methods often rely on manual records and regular reports, which is not only inefficient, but also difficult to capture subtle changes and potential risks in the construction process. At the same time, since construction data usually comes from multiple dimensions, including construction resources, construction progress, and external environment (such as weather conditions), how to effectively integrate and utilize these multi-dimensional data has become a technical problem that needs to be solved urgently.

[0003] However, existing data processing and analysis technologies are often limited to single dimensions or simple linear relationship analysis, which makes it difficult to fully reflect the complexity and dynamics of the construction process. In addition, the prediction of future construction data mainly relies on statistical models and economic theories, such as linear regression models, time series analysis (such as ARIMA models) and simple empirical models. Traditional methods cannot accurately reflect the nonlinear relationships and complex interactions of multiple factors in real estate construction projects. For example, daily material prices, labor costs, weather and construction progress have complex effects on overall costs and project progress. Traditional models are difficult to effectively cope with such a changing environment.

[0004] Therefore, traditional forecasting methods often only focus on single-dimensional data such as construction progress or resource input, ignoring the connection between multiple dimensions. The accuracy of the forecasting results cannot meet the requirements, resulting in risks in the construction monitoring process of real estate projects. Summary of the invention

[0005] One or more embodiments of the present specification provide a TimesNet model real estate construction monitoring method, equipment and medium for solving the following technical problems: Traditional prediction methods often only focus on single-dimensional data such as construction progress or resource input, ignoring the connection between multiple dimensions. The accuracy of the prediction results cannot meet the requirements, resulting in risks in the construction monitoring process of real estate projects.

[0006] One or more embodiments of this specification adopt the following technical solutions: One or more embodiments of the present specification provide a real estate construction monitoring method based on a TimesNet model, the method comprising: obtaining multi-dimensional construction time series data of a real estate construction project, wherein the multi-dimensional construction time series data comprises construction resource time series data, construction progress time series data and weather status time series data, and the construction resources comprise material resource prices, material resource quantities, human resource costs and human resource inputs; using a graph attention network to perform feature extraction on the multi-dimensional construction time series data to determine the multi-dimensional fusion construction features corresponding to the multi-dimensional construction data; determining, through a pre-trained TimesNet model, multi-dimensional predicted construction data within a corresponding predicted time length according to the multi-dimensional fusion construction features corresponding to the multi-dimensional construction data, wherein the multi-dimensional predicted construction data comprises predicted construction resource data and predicted construction progress data, and the predicted construction predicted resource data comprises predicted material resource data and predicted human resource data; performing risk monitoring on the real estate construction project according to the multi-dimensional predicted construction data, and determining the construction risk information of the real estate construction project, so as to perform risk management and control on the real estate construction project through the construction risk information.

[0007] One or more embodiments of this specification provide a TimesNet model real estate construction monitoring device, including: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the above method.

[0008] One or more embodiments of the present specification provide a non-volatile computer storage medium storing computer executable instructions, wherein the computer executable instructions are configured to execute the above method.

[0009] At least one of the above technical solutions adopted in the embodiments of this specification can achieve the following beneficial effects: through the above technical solution, by collecting multi-dimensional construction time series data, including construction resources, construction progress, weather conditions and other aspects, it can more comprehensively reflect the actual situation of the construction project, reduce information loss and deviation, and improve the accuracy of data processing; using the graph attention network to extract features from the multi-dimensional construction time series data, it can automatically learn the complex relationships and patterns in the data, and extract features that are more critical to prediction and construction risk monitoring. The attention mechanism of GAT enables the model to focus on more important nodes and connections, thereby improving the efficiency and accuracy of feature extraction; through the pre-trained TimesNet model, predictions are made based on multi-dimensional fusion construction features, which can generate The multi-dimensional predicted construction data within the corresponding prediction time length not only takes into account the changing trend of historical data, but also incorporates the current construction status and external environmental factors. Therefore, it can more accurately predict future construction conditions. At the same time, since the model is pre-trained, the prediction can be completed in a shorter time, improving the real-time nature of the prediction. According to the multi-dimensional predicted construction data, risk monitoring of real estate construction projects can be carried out, which can automatically identify and evaluate potential risks in the construction process. Intelligent risk monitoring methods can timely discover and warn of risk events, provide timely and accurate decision-making support, and the automated monitoring process also reduces the burden of manual monitoring and improves work efficiency. Through the acquisition and analysis of construction risk information, corresponding risk management and control strategies can be formulated for different risk types and degrees. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions in the embodiments of this specification or the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art description. Obviously, the drawings described below are only some embodiments recorded in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. In the drawings: Figure 1 A flowchart of a TimesNet model real estate construction monitoring method provided in an embodiment of this specification; Figure 2 A structural schematic diagram of a TimesNet model real estate construction monitoring device provided in an embodiment of this specification. DETAILED DESCRIPTION

[0011] In order to enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below in conjunction with the drawings in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on the embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this specification.

[0012] The embodiments of this specification provide a real estate construction monitoring method based on the TimesNet model. It should be noted that the execution subject in the embodiments of this specification can be a server or any device with data processing capabilities. Figure 1 A flowchart of a TimesNet model real estate construction monitoring method provided in an embodiment of this specification is shown in FIG. Figure 1 As shown, it mainly includes the following steps: Step S101, obtaining multi-dimensional construction time series data of a real estate construction project.

[0013] The multi-dimensional construction time series data includes construction resource time series data, construction progress time series data and weather status time series data. The construction resources include material resource prices, material resource quantities, human resource costs and human resource inputs. In one embodiment of the present specification, multi-dimensional construction data of a real estate construction project is obtained, and the multi-dimensional construction time series data includes construction resource time series data, construction progress time series data, and weather status time series data, and the construction resources include material resource prices, material resource quantities, human resource costs, and human resource inputs. When obtaining construction resource time series data, material price information is obtained by subscribing to industry data services or regularly from suppliers, and the price of each material at different time points is recorded using an API interface or manual entry. Use a material management system (such as ERP, WMS, etc.) to record material warehousing, outbound delivery, and inventory status, and update material quantity data in real time. Collect employee payroll, overtime pay, welfare, and other cost information, and human resource costs can be grouped by department or project. Use a work time management system or manual recording method to count daily human resource inputs, including the number of personnel, types of work, and working hours. According to the project plan, set key milestones and specific tasks, track progress through project management software (such as Primavera P6, Microsoft Project) or a custom system, and obtain construction progress time series data. Use public weather data APIs (such as the Meteorological Bureau, Weather API, etc.) to obtain real-time and historical weather data for the construction site, including temperature, humidity, precipitation, wind speed, etc.

[0014] Step S102: Using a graph attention network, feature extraction is performed on the multi-dimensional construction time series data to determine the multi-dimensional fusion construction features corresponding to the multi-dimensional construction data.

[0015] A graph attention network is used to extract features from the multidimensional construction time series data to determine the multidimensional fused construction features corresponding to the multidimensional construction data, specifically including: preprocessing the multidimensional construction time series data to determine the specified multidimensional construction time series data; constructing graph structure data according to the specified multidimensional construction time series data, wherein the graph structure data includes multiple node structures and edge structures corresponding to two nodes, the node structure represents multiple construction dimensions of the construction data, and the edge structure represents the correlation between the nodes; the graph attention network performs weight learning through a self-attention mechanism to determine the weights between multiple nodes, so as to aggregate the features of the specified neighbor nodes according to the weights between the multiple nodes, determine the node features of each node, and determine the dimensional feature data of each construction dimension; a multi-level weighting mechanism is used to perform feature fusion on the dimensional feature data corresponding to multiple construction dimensions to determine the multidimensional fused construction features corresponding to the multidimensional construction data.

[0016] In one embodiment of the present specification, the collected data is cleaned, standardized, processed for outliers, and filled with missing data to ensure that the data quality meets the needs of further analysis. Through this series of preprocessing measures, the system can eliminate noise in the data, correct errors, fill in missing values, and unify the data format, thereby providing high-quality data support for subsequent feature extraction and model training.

[0017] The Graph Attention Network (GAT) model is used to extract deep features from multi-dimensional construction time series data, focusing on the dependencies and spatiotemporal dynamics between variables, extracting and fusing features from multi-dimensional data, and deeply exploring the potential dependencies in the data, thereby improving the accuracy of prediction. In order to fully explore and utilize the rich information of multimodal data, deep feature learning is performed on the input features: the GAT model uses the attention mechanism to perform deep feature extraction and modeling on these data. The key is to be able to dynamically adjust the weight of each variable. For example, in the prediction process, weather changes, labor costs, construction progress, etc. may have different degrees of impact. The GAT model can learn the dependencies between variables and assign different weights to ensure that the role of important variables is fully reflected. The attention mechanism of the GAT model not only focuses on the static relationship between variables, but also takes into account the changes in spatiotemporal dynamics. Specifically, the long-term and short-term dependencies in the data can be captured by GAT. Especially when dealing with time series data with periodicity and complex nonlinear trends, the GAT model can effectively identify long-term trends and short-term fluctuations. This feature extraction capability enables the system to deeply understand the hidden laws in the real estate market and improve the accuracy and robustness of predictions.

[0018] After feature extraction is completed, the features extracted from different dimensions are weighted and fused. The feature fusion process adopts a multi-level weighting mechanism to uniformly process feature data from different sources to form a high-dimensional feature representation. This feature representation not only has the comprehensive advantages of data in each dimension, but also can provide higher-quality input for the subsequent TimesNet model, further enhancing the model's prediction performance. In this way, the feature extraction and fusion module provides a strong data support and analysis basis for the system's financial risk prediction, ensuring that the system has high-precision prediction capabilities when dealing with complex market environments.

[0019] Step S103, through the pre-trained TimesNet model, according to the multi-dimensional fusion construction features corresponding to the multi-dimensional construction data, determine the multi-dimensional predicted construction data within the corresponding predicted time length.

[0020] The multi-dimensional predicted construction data includes predicted construction resource data and predicted construction progress data, and the predicted construction predicted resource data includes predicted material resource data and predicted human resource data; Before determining the multi-dimensional predicted construction data within the corresponding predicted time length according to the multi-dimensional fusion construction features corresponding to the multi-dimensional construction data through the pre-trained TimesNet model, the method also includes: collecting multiple historical construction data to construct a training data set, wherein the training data set includes multiple historical construction time features corresponding to the complete construction cycle and multi-dimensional historical construction feature data corresponding to each of the historical construction time features, and the multi-dimensional historical construction feature data includes a material resource demand dimension, a material resource market cost dimension, and a weather dimension; constructing an initial TimesNet model, wherein the initial TimesNet model includes an input layer, a hidden layer, and an output layer; inputting the training data set into the initial TimesNet model, training the initial TimesNet model, and determining the trained TimesNet model.

[0021] Collecting multiple historical construction data to construct a training data set specifically includes: obtaining multi-dimensional historical construction time series data corresponding to multiple complete construction cycles, and determining the time period information of each complete construction cycle, wherein the multi-dimensional historical construction time series data includes historical time data and multi-dimensional historical construction data; determining the occupied time ratio of the historical time data within the complete construction cycle according to the historical time data and the time period information corresponding to the complete construction cycle; quantifying the historical time data through the occupied time ratio, determining the historical construction time characteristics, and constructing the training data set based on the historical construction time characteristics and the multi-dimensional historical construction data.

[0022] In one embodiment of the present specification, multiple historical construction data are collected to construct a training data set, and the training data set is used to train the model. In this process, the training data set includes multiple historical construction time features corresponding to the complete construction cycle and multi-dimensional historical construction feature data corresponding to each of the historical construction time features. The multi-dimensional historical construction feature data includes material resource demand dimension, material resource market cost dimension and weather dimension. It should be noted that, in general, the collected time data is specific time information. Since the role of construction time in different construction cycles is different in different real estate projects, the time information can be quantified. Determine the time period information of each complete construction cycle, for example, the complete cycle is 500 days. Determine the number of days that have been constructed according to the historical time data, and determine the ratio of the number of days to the time period information corresponding to the complete construction cycle as the occupied time ratio. Through the occupied time ratio, the historical time data is tabled to determine the historical construction time characteristics, so as to construct a training data set based on the historical construction time characteristics and multi-dimensional historical construction data. After the multi-dimensional construction time series data is collected, if the indicator quantization method is used in the model training stage, the multi-dimensional construction time series data needs to be quantified and converted into the corresponding time feature indicators. The TimesNet model is used to process complex time series data and predict process data in real estate construction. By leveraging the powerful capabilities of the TimesNet model, it can efficiently capture long-term dependencies and short-term dynamic changes in time series data, thereby improving prediction accuracy and stability.

[0023] In one embodiment of the present specification, the deep fusion features processed by the graph attention network (GAT) are used as input. These features have undergone pre-processing steps such as data cleaning, standardization, outlier processing and missing value filling to ensure the quality and consistency of the data. The model architecture includes an input layer, a hidden layer and an output layer, wherein the hidden layer uses deep learning techniques (such as convolutional layers, recurrent neural network layers, etc.) to capture patterns in the data, and can effectively process and predict important events with long intervals and delays in time series data. During the training process, the model parameters are updated by setting initial parameters, defining loss functions (such as mean square error) and selecting optimization algorithms (such as Adam) to minimize the loss function, and select L1 or L2 regularization techniques to prevent overfitting, and select the best hyperparameter combination through cross-validation. After the training is completed, the model can be used to predict the construction process data in the construction phase, such as real-time prediction of daily material costs, labor costs and project progress, and provide optimization suggestions. The model also applies the Fast Fourier Transform (FFT) and introduces a multi-periodic perspective to model the changing patterns in the time series, so that the model can more effectively capture the periodic and non-periodic features in the data. The folding mechanism enables simple 2D convolution to extract richer features when processing 1D original sequences. The model can adaptively extract the period length from the data. Even if the periods extracted from different batches of data may be different, if the periodicity of the data itself is relatively fixed, the period lengths calculated from different batches will be similar. Through the above steps, the TimesNet model construction module can effectively predict the real estate construction process.

[0024] Step S104, based on the multi-dimensional predicted construction data, risk monitoring is performed on the real estate construction project to determine the construction risk information of the real estate construction project, so as to perform risk management and control on the real estate construction project through the construction risk information.

[0025] Based on the multi-dimensional predicted construction data, risk monitoring is performed on the real estate construction project to determine the construction risk information of the real estate construction project, specifically including: obtaining pre-set construction plan information, wherein the construction plan information includes construction progress plan data, construction resource supply plan data and construction cost plan data; obtaining predicted construction resource data and predicted construction progress data in the multi-dimensional predicted construction data, wherein the predicted construction resource data includes predicted construction resource demand data and predicted construction resource cost data corresponding to each predicted timestamp, and the predicted construction progress data includes predicted construction progress nodes corresponding to each predicted timestamp; based on the multi-dimensional predicted construction data and the construction plan information, multi-dimensional risk monitoring is performed on the real estate construction project to determine the construction risk information of the real estate construction project, wherein the multi-dimensional risk monitoring includes a construction resource cost dimension, a construction progress dimension and a construction risk resource dimension.

[0026] Based on the multi-dimensional predicted construction data and the construction plan information, multi-dimensional risk monitoring is performed on the real estate construction project to determine the construction risk information of the real estate construction project, specifically including: determining the predicted construction resource input data corresponding to each predicted timestamp through the predicted construction resource cost data and the predicted construction resource demand data corresponding to each predicted timestamp; matching the node resource planned input data corresponding to the same construction progress node in the construction cost plan data based on the predicted construction progress data corresponding to the predicted timestamp, comparing the construction resource predicted input data with the node resource planned input data, and determining the resource input risk information corresponding to each predicted timestamp; matching the corresponding planned construction time information in the construction progress plan data corresponding to the construction plan information according to the predicted construction progress node in the predicted construction progress data corresponding to the predicted timestamp; The predicted timestamp is compared with the planned construction time information, and the construction progress of the real estate construction project is monitored to determine whether the predicted timestamp and the planned construction time information are within the preset time deviation range; if not, based on the relative time relationship between the predicted timestamp and the planned construction time information, determine whether the predicted timestamp has a construction progress risk, so as to determine the construction progress risk information corresponding to each predicted timestamp; based on the predicted construction resource demand data and the construction resource supply plan data, determine the construction risk resource information corresponding to each predicted timestamp, wherein the construction risk resource information includes a risk type and a risk resource type, and the risk type includes any one of a construction resource backlog risk and a construction resource shortage risk; through the resource input risk information, the construction progress risk information and the construction risk resource information, determine the construction risk information corresponding to each predicted timestamp of the real estate construction project.

[0027] In one embodiment of the present specification, the predicted construction resource cost data and predicted construction resource demand data corresponding to each predicted timestamp are used to determine the predicted construction resource input data corresponding to each predicted timestamp. It should be noted that the predicted input cost of each construction resource should be calculated separately for different types of construction resources, and the embodiments of the present specification are discussed with the same construction resource. Based on the predicted construction progress data corresponding to the predicted timestamp, the node resource planned input data corresponding to the same construction progress node is matched in the construction cost plan data, and the node resource planned input data is used as the standard. The construction resource predicted input data is compared with the node resource planned input data to determine the resource input risk information corresponding to each predicted timestamp, and the resource input excess data compared with the construction resource predicted input data and the node resource planned input data is calculated. The resource input excess data is used as the resource input risk information. If the construction resource predicted input data is not greater than the node resource planned input data, it means that there is no risk of resource input excess.

[0028] According to the predicted construction progress node in the predicted construction progress data corresponding to the predicted timestamp, the corresponding planned construction time information is matched in the construction progress plan data corresponding to the construction plan information; the predicted timestamp is compared with the planned construction time information, the construction progress of the real estate construction project is monitored, and it is determined whether the time difference between the predicted timestamp and the planned construction time information is within the preset time deviation range, and the time deviation range here is set according to the empirical data. If it is not within the preset time deviation range, then according to the time relative relationship between the predicted timestamp and the planned construction time information, the time relative relationship includes three situations: the predicted timestamp is before the planned construction time information, the predicted timestamp coincides with the planned construction time information, and the predicted timestamp is after the planned construction time information. According to the time relative relationship, it is determined whether the predicted timestamp has a construction progress risk, so as to determine the construction progress risk information corresponding to each predicted timestamp. It should be noted that if the predicted timestamp is before the planned construction time information or the predicted timestamp coincides with the planned construction time information, it means that there is no construction progress risk. If the predicted timestamp is after the planned construction time information, that is, the predicted timestamp is later than the planned construction time, it means that there is a construction progress risk.

[0029] Based on the predicted construction resource demand data and the construction resource supply plan data, the demand quantity and supply data are compared to determine the construction risk resource information corresponding to each predicted timestamp, wherein the construction risk resource information includes a risk type and a risk resource type, wherein the risk type includes any one of a construction resource backlog risk and a construction resource shortage risk, and the risk resource type is used to indicate a specific construction resource with a risk. The resource input risk information, the construction progress risk information and the construction risk resource information are integrated to determine the construction risk information corresponding to each predicted timestamp of the real estate construction project.

[0030] Through the construction risk information, risk control is performed on the real estate construction project, specifically including: obtaining the construction risk information, wherein the construction risk information includes resource input risk information, construction progress risk information and construction risk resource information corresponding to each prediction timestamp; based on the resource input risk information, construction progress risk information and construction risk resource information corresponding to each prediction timestamp, risk quantification is performed on each prediction timestamp to determine the construction risk quantification index of each prediction timestamp; according to the construction risk quantification index of each prediction timestamp, control time node information and corresponding risk control strategy are determined.

[0031] In one embodiment of the present specification, based on the TimesNet model prediction tool, forecast data such as resource input, construction progress and weather conditions corresponding to each forecast timestamp (such as daily, weekly or monthly) are generated. Based on these forecast data, the resource input risk information (such as resource shortage, cost overrun, etc.), construction progress risk information (such as delay, early completion, etc.) and construction risk resource information (such as shortage of key resources, unreasonable resource allocation, etc.) of each forecast timestamp are analyzed and extracted. The resource input risk, construction progress risk and construction risk resource information of each forecast timestamp are quantitatively evaluated, and the construction risk quantitative index of each forecast timestamp is calculated by setting a certain risk indicator system and quantitative method. According to the level of the construction risk quantitative index, the control time node is identified. It should be noted that since the construction progress is greatly affected by the construction time, when determining the control time point, the construction risk quantitative index of each forecast timestamp can be judged by referring to the index threshold, and the timestamp with the earliest time is determined as the control time point among the multiple specified forecast timestamps higher than the reference index threshold. Generally speaking, if a risk has occurred at a certain time point, the possibility of construction risks at subsequent time nodes will increase. Therefore, the earliest timestamp is selected as the control time point. For this control time point, a corresponding risk control strategy is formulated, including resource allocation, schedule adjustment, and strengthening of safety measures, to ensure the smooth progress of the construction project. For the control time point node, a specific risk control strategy is formulated for the specific risk type (resource input risk, construction schedule risk, construction risk resource information). For example, for resource input risk, strategies may include advance procurement, finding alternative suppliers, optimizing resource utilization efficiency, etc.; for construction schedule risk, strategies may include increasing manpower or machinery input, adjusting the construction sequence, working overtime, etc.; for construction risk resource information, strategies may include reserving key resources in advance and establishing an emergency response mechanism with suppliers.

[0032] In one embodiment of the present specification, based on the resource input risk information, construction progress risk information and construction risk resource information corresponding to each prediction time stamp, each prediction time stamp is risk quantified, and the construction risk quantification index of each prediction time stamp is determined, specifically including: determining the resource input risk information corresponding to each prediction time stamp, wherein the resource input risk information includes the resource input excess data between the construction resource prediction input data and the node resource plan input data; generating the resource input risk index by the ratio of the resource input excess data to the node resource plan input data corresponding to the prediction time stamp. Determine the construction progress risk information corresponding to each prediction time stamp, wherein the construction progress risk information includes the construction progress delay time information, and determine the construction progress risk delay index based on the ratio of the construction progress delay time information and the maximum deviation threshold in the preset time deviation range. For example, the construction progress delay time information is 3 days, the preset time deviation range is two days before and after, and the maximum deviation threshold in the time deviation range is 2. The ratio of 3 to 2 is calculated and determined as the construction progress risk delay index. Determine the construction risk resource information corresponding to each prediction timestamp, wherein the construction risk resource information includes the number of resource shortages, so as to determine the construction resource risk assessment index based on the ratio of the number of resource shortages to the construction resource supply plan data.

[0033] According to the resource input risk index, the construction progress risk delay index and the construction resource risk assessment index, the construction risk quantitative index of each prediction time stamp is determined. It should be noted that the weighted average method can be used here. Assume that Rt represents the resource input risk index of the t-th prediction time stamp, Dt represents the construction progress risk delay index of the t-th prediction time stamp, and At represents the construction resource risk assessment index of the t-th prediction time stamp. According to the project characteristics and the focus of risk management, corresponding weights wR, wD and wA are assigned to Rt, Dt and At, and the weights should satisfy wR+wD+wA=1. The construction risk quantitative index Qt of the t-th prediction time stamp is calculated using the weighted average method: Qt=wR⋅Rt+wD⋅Dt+wA⋅At, and Qt is standardized so that it is within a specific numerical range (such as 0 to 100) for easy comparison and display. The higher the value of Qt, the greater the construction risk of the t-th prediction time stamp. Based on the value of Qt, we can determine which timestamps are the focus of risk management and control, and formulate corresponding risk response strategies.

[0034] Through the above technical solution, by collecting multi-dimensional construction time series data, including construction resources, construction progress, weather conditions and other aspects, it can more comprehensively reflect the actual situation of the construction project, reduce information loss and deviation, and improve the accuracy of data processing; using the graph attention network to extract features from multi-dimensional construction time series data, it can automatically learn the complex relationships and patterns in the data and extract features that are more critical to prediction and construction risk monitoring. The attention mechanism of GAT enables the model to focus on more important nodes and connections, thereby improving the efficiency and accuracy of feature extraction; through the pre-trained TimesNet model, predictions are made based on multi-dimensional fusion construction features, which can generate multi-dimensional predictions within the corresponding prediction time length. Construction data not only takes into account the changing trends of historical data, but also incorporates the current construction status and external environmental factors. Therefore, it can more accurately predict future construction conditions. At the same time, since the model is pre-trained, the prediction can be completed in a shorter time, improving the real-time nature of the prediction. Based on multi-dimensional predicted construction data, risk monitoring of real estate construction projects can automatically identify and evaluate potential risks in the construction process. Intelligent risk monitoring methods can promptly detect and warn of risk events, provide timely and accurate decision-making support, and the automated monitoring process also reduces the burden of manual monitoring and improves work efficiency. Through the acquisition and analysis of construction risk information, corresponding risk management and control strategies can be formulated for different risk types and degrees.

[0035] The present specification also provides a TimesNet model of real estate construction monitoring equipment, such as Figure 2 As shown, the device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the above method.

[0036] The embodiments of the present specification also provide a non-volatile computer storage medium storing computer executable instructions, wherein the computer executable instructions are configured to execute the above method.

[0037] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device, equipment, and non-volatile computer storage medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0038] The above is a description of a specific embodiment of the specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in an order different from that in the embodiments and still achieve the desired results. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0039] The devices and media provided in the embodiments of this specification correspond one-to-one to the methods. Therefore, the devices and media also have similar beneficial technical effects as the corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the devices and media will not be repeated here.

[0040] Those skilled in the art will appreciate that the embodiments of this specification may be provided as methods, systems, or computer program products. Therefore, this specification may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, this specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0041] This specification is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of this specification. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0042] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0043] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0044] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0045] Memory may include non-permanent storage in a computer-readable medium, in the form of random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0046] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0047] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0048] The above description is only one or more embodiments of this specification and is not intended to limit this specification. For those skilled in the art, one or more embodiments of this specification may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of one or more embodiments of this specification shall be included in the scope of the claims of this specification.

Claims

1. A TimesNet model real estate construction monitoring method, characterized in that: The method comprises: Acquire multi-dimensional construction time series data of a real estate construction project, wherein the multi-dimensional construction time series data includes construction resource time series data, construction progress time series data, and weather status time series data, and the construction resources include material resource prices, material resource quantities, human resource costs, and human resource inputs; Using a graph attention network, extracting features from the multi-dimensional construction time series data to determine multi-dimensional fusion construction features corresponding to the multi-dimensional construction data; Through the pre-trained TimesNet model, according to the multi-dimensional fusion construction features corresponding to the multi-dimensional construction data, the multi-dimensional predicted construction data within the corresponding predicted time length is determined, wherein the multi-dimensional predicted construction data includes predicted construction resource data and predicted construction progress data, and the predicted construction predicted resource data includes predicted material resource data and predicted human resource data; Based on the multi-dimensional predicted construction data, risk monitoring is performed on the real estate construction project to determine the construction risk information of the real estate construction project, so as to perform risk management and control on the real estate construction project through the construction risk information.

2. According to the TimesNet model of claim 1, the real estate construction monitoring method is characterized in that: Using a graph attention network, feature extraction is performed on the multi-dimensional construction time series data to determine the multi-dimensional fusion construction features corresponding to the multi-dimensional construction data, specifically including: Preprocessing the multi-dimensional construction time series data to determine the specified multi-dimensional construction time series data; Constructing graph structure data according to the specified multi-dimensional construction time series data, wherein the graph structure data includes multiple node structures and edge structures corresponding to two nodes, the node structure represents multiple construction dimensions of the construction data, and the edge structure represents the correlation between the nodes; The graph attention network performs weight learning through a self-attention mechanism to determine weights between the multiple nodes, so as to aggregate features of designated neighbor nodes according to the weights between the multiple nodes, determine node features of each node, and determine dimensional feature data of each construction dimension; A multi-level weighting mechanism is adopted to perform feature fusion on the dimensional feature data corresponding to the multiple construction dimensions, and determine the multi-dimensional fused construction features corresponding to the multi-dimensional construction data.

3. The real estate construction monitoring method of a TimesNet model according to claim 1, characterized in that: Before determining the multi-dimensional predicted construction data within the corresponding predicted time length by using the pre-trained TimesNet model and according to the multi-dimensional fusion construction features corresponding to the multi-dimensional construction data, the method further includes: Collecting multiple historical construction data to construct a training data set, wherein the training data set includes multiple historical construction time features corresponding to a complete construction cycle and multi-dimensional historical construction feature data corresponding to each of the historical construction time features, and the multi-dimensional historical construction feature data includes a material resource demand dimension, a material resource market cost dimension, and a weather dimension; Constructing an initial TimesNet model, wherein the initial TimesNet model includes an input layer, a hidden layer, and an output layer; The training data set is input into the initial TimesNet model, the initial TimesNet model is trained, and the trained TimesNet model is determined.

4. The real estate construction monitoring method of a TimesNet model according to claim 1, characterized in that: According to the multi-dimensional predicted construction data, risk monitoring is performed on the real estate construction project to determine the construction risk information of the real estate construction project, specifically including: Acquire preset construction plan information, wherein the construction plan information includes construction progress plan data, construction resource supply plan data and construction cost plan data; Acquire predicted construction resource data and predicted construction progress data in the multi-dimensional predicted construction data, wherein the predicted construction resource data includes predicted construction resource demand data and predicted construction resource cost data corresponding to each predicted timestamp, and the predicted construction progress data includes predicted construction progress nodes corresponding to each predicted timestamp; Based on the multi-dimensional predicted construction data and the construction plan information, multi-dimensional risk monitoring is performed on the real estate construction project to determine the construction risk information of the real estate construction project, wherein the multi-dimensional risk monitoring includes the construction resource cost dimension, the construction progress dimension and the construction risk resource dimension.

5. The real estate construction monitoring method of the TimesNet model according to claim 4 is characterized in that: Performing multi-dimensional risk monitoring on the real estate construction project according to the multi-dimensional predicted construction data and the construction plan information to determine the construction risk information of the real estate construction project specifically includes: Determine the construction resource forecast input data corresponding to each forecast time stamp through the forecast construction resource cost data and the forecast construction resource demand data corresponding to each forecast time stamp; Based on the predicted construction progress data corresponding to the predicted timestamp, the node resource plan input data corresponding to the same construction progress node is matched in the construction cost plan data, the construction resource predicted input data and the node resource plan input data are compared to determine the resource input risk information corresponding to each predicted timestamp; According to the predicted construction progress node in the predicted construction progress data corresponding to the predicted timestamp, matching the corresponding planned construction time information in the construction progress plan data corresponding to the construction plan information; Monitoring the construction progress of the real estate construction project by using the predicted timestamp and the planned construction time information, and determining whether the predicted timestamp and the planned construction time information are within a preset time deviation range; If not, judging whether the predicted timestamp has a construction progress risk based on the relative time relationship between the predicted timestamp and the planned construction time information, so as to determine the construction progress risk information corresponding to each predicted timestamp; Based on the predicted construction resource demand data and the construction resource supply plan data, determining the construction risk resource information corresponding to each predicted timestamp, wherein the construction risk resource information includes a risk type and a risk resource type, and the risk type includes any one of a construction resource backlog risk and a construction resource shortage risk; The construction risk information corresponding to each predicted timestamp of the real estate construction project is determined through the resource investment risk information, the construction progress risk information and the construction risk resource information.

6. The real estate construction monitoring method of a TimesNet model according to claim 1, characterized in that: The risk management of the real estate construction project is carried out through the construction risk information, specifically including: Acquire the construction risk information, wherein the construction risk information includes resource input risk information, construction progress risk information and construction risk resource information corresponding to each prediction timestamp; Based on the resource investment risk information, construction progress risk information and construction risk resource information corresponding to each of the prediction timestamps, risk quantification is performed on each of the prediction timestamps to determine the construction risk quantification index of each of the prediction timestamps; According to the construction risk quantification index of each predicted timestamp, the control time node information and the corresponding risk control strategy are determined.

7. The real estate construction monitoring method of the TimesNet model according to claim 6 is characterized in that: Based on the resource investment risk information, construction progress risk information and construction risk resource information corresponding to each prediction time stamp, risk quantification is performed on each prediction time stamp to determine the construction risk quantification index of each prediction time stamp, specifically including: Determine resource input risk information corresponding to each of the predicted timestamps, wherein the resource input risk information includes resource input over-standard data between the construction resource prediction input data and the node resource plan input data; Generate a resource investment risk indicator by using the ratio of the resource investment excess data to the node resource plan investment data corresponding to the prediction timestamp; Determine the construction progress risk information corresponding to each of the predicted timestamps, wherein the construction progress risk information includes construction progress delay time information, and determine the construction progress risk delay index based on the ratio of the construction progress delay time information to a maximum deviation threshold in a preset time deviation range; Determine the construction risk resource information corresponding to each of the prediction timestamps, wherein the construction risk resource information includes the number of resource shortages, so as to determine the construction resource risk assessment index based on the ratio of the number of resource shortages to the construction resource supply plan data; The construction risk quantification index of each predicted timestamp is determined according to the resource input risk index, the construction progress risk delay index and the construction resource risk assessment index.

8. The real estate construction monitoring method of the TimesNet model according to claim 3 is characterized in that: Collect multiple historical construction data to build a training dataset, including: Acquire multi-dimensional historical construction time series data corresponding to a plurality of complete construction cycles, and determine time period information of each of the complete construction cycles, wherein the multi-dimensional historical construction time series data includes historical time data and multi-dimensional historical construction data; Determine the occupied time ratio of the historical time data within the complete construction period according to the historical time data and the time period information corresponding to the complete construction period; The historical time data is quantified by the occupied time ratio to determine the historical construction time characteristics, so as to construct the training data set based on the historical construction time characteristics and the multi-dimensional historical construction data.

9. A TimesNet model real estate construction monitoring device, characterized in that: The device comprises: at least one processor; and, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can perform the method according to any one of claims 1 to 8.

10. A non-volatile computer storage medium storing computer executable instructions, characterized in that: The computer executable instructions are configured to execute the method according to any one of claims 1 to 8.

Images